4; but in practice they have always adopted definition No. 3. The consequence is, that in literally numberless cases (particularly in the vegetable kingdom) "specific characters" are assumed to be hereditary characters merely because systematic naturalists have bestowed a specific name on the form which presents them. Nor is this all. For, conversely, even when it is known that constant morphological characters are unquestionably hereditary characters, if they happen to present but small degrees of divergence from those of allied forms, then the form which presents them is not ranked as a species, but as a constant variety. In other words, when definitions 3 and 4 are found to clash, it is not 4, but 3, that is followed. In short, even up to the present time, systematic naturalists play fast and loose with the criterion of Heredity to such an extent, that, as above observed, it has been rendered wellnigh worthless in fact, whatever may be thought of it in theory. Now, unless all this can be denied, what is the use of representing that a species is distinguished from a variety--"climatic" or otherwise--by the fact that its constituent individuals "reproduce their like"? We are not here engaged on any abstract question of what might have been the best principles of specific distinction for naturalists to have adopted. We are engaged on the practical question of the principles which they actually have adopted. And of these principles the reproduction of like by like, under all circumstances of environment, has been virtually ignored. (B.) In the second place, supposing that the criterion of Heredity had been as universally and as rigidly employed by our systematists in their work of constructing species as it has been but occasionally and loosely employed, could it be said that even then a basis would have been furnished for the doctrine that all specific characters must necessarily be useful characters? Obviously not, and for the following reasons. It is admitted that climatic characters are not necessarily--or even generally--useful characters. Consequently, if there be any reason for believing that climatic characters may become in time hereditary characters, the doctrine in question would collapse, even supposing that all specific types were to be re-constituted on a basis of experimental inquiry, for the purpose of ascertaining which of them conform to the test of Heredity. Now there are very good reasons for believing that climatic characters not unfrequently do become hereditary characters; and it was mainly in view of those reasons that I deemed it worth while to devote so much space in the preceding chapter to the facts of climatic variation. I will now state the reasons in question under two different lines of argument.